Autosegmental Representations in Zigula and Shambaa

This chapter provides computational analyses of two common patterns in tone, unbounded tone shift and unbounded tone spreading, using autosegmental representations (Goldsmith, 1976). These patterns, exemplified here by data from two Bantu languages, illustrate two common characteristics of tone (see Yip, 2002). Unbounded tone shift in Zigula (Kenstowicz and Kisseberth, 1990) exemplifies the ‘mobility’ of tonal units, or their ability to move long distances. Unbounded tone spreading, exemplified by data from Shambaa (Odden, 1982), illustrates the ability of a single tonal unit to be associated to multiple vowels. This chapter shows how MSO-definable transductions over autosegmental representations can insightfully capture these phenomena. We first turn to the distribution of tones in Zigula verbs (also known as Zigua or Chizigula; Kenstowicz and Kisseberth, 1990), and argue that the correct generalization is that a single underlying tone shifts to the penultimate vowel. Zigula verb roots come in two flavors: toned and toneless, as can be seen in the infinitive forms in Table 1. The following data are due to Kenstowicz and Kisseberth (1990). Surface high tones are marked with an acute accent on the vowel ([á]); low-toned vowels are unmarked. In these and all examples in this chapter, the data have been converted to IPA from their original sources.